Fairer Decisions, Better Health
for All: Health Equity and CostEffectiveness Analysis
Richard Cookson, Andrew Mirelman,
Miqdad Asaria, Bryony Dawkins,
Susan Griffin
CHE Research Paper 135
Fairer decisions, better health for all: Health equity and costeffectiveness analysis
1
Richard Cookson
Andrew Mirelman
1
Miqdad Asaria
2
Bryony Dawkins
1
Susan Griffin
1
1
Centre for Health Economics, University of York, York, UK.
Academic Unit of Health Economics, Leeds Institute of Health Sciences, University of Leeds,
UK
2
September 2016
Background to series
CHE Discussion Papers (DPs) began publication in 1983 as a means of making current
research material more widely available to health economists and other potential users. So
as to speed up the dissemination process, papers were originally published by CHE and
distributed by post to a worldwide readership.
The CHE Research Paper series takes over that function and provides access to current
research output via web-based publication, although hard copy will continue to be available
(but subject to charge).
Acknowledgements
Richard Cookson and Miqdad Asaria are supported by the National Institute for Health
Research (Senior Research Fellowship, Dr Richard Cookson, SRF-2013-06-015). The views
expressed in this publication are those of the authors and not necessarily those of the NHS,
the National Institute for Health Research or the Department of Health.
For helpful comments the authors would like to thank Gwyn Bevan, Alan Brennan, Simon
Capewell, Brendan Collins, Kalipso Chalkidou, Tony Culyer, Brian Ferguson, Amanda
Glassman, John Holmes, Carleigh Krubiner, Ryan Li, Martin O’Flaherty, Ole Norheim, Marc
Suhrcke, Aki Tsuchiya, Stephane Verguet and Adam Wagstaff; though our errors and
opinions are our own.
Further copies
Copies of this paper are freely available to download from the CHE website
www.york.ac.uk/che/publications/ Access to downloaded material is provided on the
understanding that it is intended for personal use. Copies of downloaded papers may be
distributed to third-parties subject to the proviso that the CHE publication source is properly
acknowledged and that such distribution is not subject to any payment.
Printed copies are available on request at a charge of £5.00 per copy. Please contact the
CHE Publications Office, email [email protected], telephone 01904 321405 for further
details.
Centre for Health Economics
Alcuin College
University of York
York, UK
www.york.ac.uk/che
© Richard Cookson, Andrew Mirelman, Miqdad Asaria, Bryony Dawkins, Susan Griffin
Fairer decisions, better health for all:health equity and cost-effectiveness analysis i
Contents
Abstract ................................................................................................................................................... ii
Executive summary ................................................................................................................................ iii
Introduction ............................................................................................................................................ 1
Part I: Concepts ....................................................................................................................................... 3
Cost-effectiveness analysis ................................................................................................................. 3
Health equity....................................................................................................................................... 4
Accounting for the social distribution of opportunity costs ............................................................... 6
Trade-offs between total health and health equity ........................................................................... 9
Measuring health equity impacts ..................................................................................................... 10
Quantifying health equity trade-offs ................................................................................................ 13
Part II: Methods .................................................................................................................................... 14
Equity evidence review ..................................................................................................................... 14
Equity impact analysis....................................................................................................................... 16
Equity constraint analysis ................................................................................................................. 18
Equity weighting analysis .................................................................................................................. 19
Conclusion ............................................................................................................................................. 22
Further reading ..................................................................................................................................... 24
Annex: Computation of the distribution of healthy life expectancy in Ethiopia .................................. 25
References ............................................................................................................................................ 28
ii CHE Research Paper 135
Abstract
This report provides a non-technical introduction to practical methods for using cost-effectiveness
analysis to address health equity concerns, with applications to low-, middle- and high-income
countries. These methods can provide information about the likely impacts of alternative health
policy decisions on inequalities in health, financial risk protection and other health-related outcomes
that may be considered unfair, allowing for the distribution of costs as well as benefits. They can
also provide information about the trade-offs that sometimes arise between improving total health
and reducing health inequalities of different kinds. We distinguish three general ways of using costeffectiveness analysis to address health equity concerns: (1) equity impact analysis, which quantifies
the distribution of costs and effects across a population by equity-relevant variables such as
socioeconomic status, ethnicity, location, gender, age and severity of illness; (2) equity constraint
analysis, which counts the cost of choosing fairer but less cost-effective options; and (3) equity
weighting analysis, which uses equity weights or parameters to explore how much concern for
equity is required to choose fairer but less cost-effective options. We hope this report will raise
awareness of the practical tools of cost-effectiveness analysis that are now available to help give
health care and public health policy makers a better understanding of who gains and who loses from
their priority setting decisions.
Further resources are available at this website:
http://www.york.ac.uk/che/research/equity/economic_evaluation/
Fairer decisions, better health for all:health equity and cost-effectiveness analysis iii
Executive summary
Introduction
Health equity has risen to prominence on policy agendas across the world, in the wake of the
universal health coverage movement and landmark international reports on social inequalities in
health and health care. Conventional cost-effectiveness analysis (CEA) can help decision makers
identify smarter investments that deliver larger health gains from limited resources. Currently,
however, most CEA studies do not provide information about who gains and who loses from policy
decisions. Both producers and users of cost-effectiveness analysis need to start paying more
attention to health equity concerns – and this report shows how.
Cost-effectiveness analysis
In the health sector, CEA usually focuses on getting the largest total health benefit for a given cost.
Standard CEA of this kind can help to inform decision making in at least four different health policy
contexts, each of which involves different though overlapping communities of decision makers,
analysts and stakeholders:
1.
2.
3.
4.
Purchasing costly new health care technologies – e.g. whether to fund imatinib for
stomach cancer, at what price, and for which patients?
Designing health care benefit packages – e.g. whether to cover diabetes in a public
health insurance plan, and if so which diabetes treatments to include?
Investing in health care infrastructure – e.g. whether to invest in a community health
worker programme, and if so how to prioritise investments in different geographical
areas?
Public health – e.g. whether to implement a sugar tax, and if so how much should it be?
CEA provides decision makers with a standard way of assessing value for money. It compares the
value of the outcomes generated by a policy intervention against those that could have been
achieved with alternative uses of the same resources, i.e. the ‘opportunity costs’. In (1), for
example, the funding could be used for other hospital-based cancer treatments – or, with more
difficulty, shifted towards primary care. In (2), there may only be enough income from premiums or
taxes to cover diabetes or dementia – but not both. In (3), the investment could instead be used to
buy hospital equipment. Lastly, in (4), a sugar tax will increase the price of sugar-sweetened
products, potentially requiring households to cut back on consumption. In each case, equity may
play an important but typically unanalysed role: standard CEA does not tell us who gains the most,
nor who bears the heaviest burden of opportunity costs.
Trade-offs between maximising total health and reducing health inequities
In ‘win-win’ cases standard CEA may be all that is needed, because the cost-effective policy that
maximises total health also reduces a particular health inequity. However, more careful analysis of
equity impacts may be needed in ‘win-lose’ cases, when a policy improves total health but harms
health equity, and in ‘lose-win’ cases, when improvement in equity comes at the expense of lower
total health.
Improving total health and health equity are often aligned, and ‘win-win’ cases are common. When
delivered equally to all social groups, for example, programmes of vaccination and primary care for
infectious diseases may be not only highly cost-effective but also more beneficial to disadvantaged
groups at higher risk of infection. But delivery is not always equal. For example, delivery costs may
be relatively high and effective healthcare coverage relatively low in poor urban areas and remote
rural areas which lack well-resourced clinics and struggle to recruit qualified medical staff. In such
iv CHE Research Paper 135
cases, delivery may be unequal unless additional resources are invested to facilitate delivery to
disadvantaged groups – resulting in trade-offs between cost-effectiveness and health equity, and
hard choices between more equal delivery versus larger total health gains.
Three ways of using cost-effectiveness analysis to address health equity concerns
(1)
(2)
(3)
Equity impact analysis. The effects of alternative policy options can be broken down by
one or more social variables. The distribution of opportunity costs may also be
presented, based on evidence and assumptions about where the funding comes from
and who would otherwise have gained the most from alternative use of scarce
resources.
Equity constraint analysis. Equity can be analysed as an ethical constraint on the
pursuit of cost-effectiveness, rather than as an outcome. The opportunity cost of
imposing this ethical constraint can be calculated as the difference in total health
benefit between the policy that offers the most total health and a more equitable policy
option.
Equity weighting analysis. When there are trade-offs between improving total health
and reducing health inequity, the implications of different value judgements about
those trade-offs can be analysed by conducting sensitivity analysis around how much
decision makers care about reducing health inequity. This can be done using ‘equity
weights’ for health benefits to people with different characteristics, or ‘equity
parameters’ quantifying the degree of concern for reducing health inequity versus
improving total health.
Alongside these analyses, one can also conduct an equity evidence review to provide useful
information for decision makers by identifying previous studies on the relevant health inequities and
the distributional impacts of similar policies that have been studied in other contexts.
Two specific methods of equity-informative cost-effectiveness analysis
We describe two specific methods of equity-informative CEA. Extended Cost-Effectiveness Analysis
(ECEA), developed by the Disease Control Priorities, 3rd edition (DCP3) project (www.dcp-3.org),
examines the social distribution of costs, health effects and financial risk protection effects.
Distributional Cost-Effectiveness Analysis (DCEA), developed by the University of York, examines the
distribution of health benefits and opportunity costs and then, if health equity trade-offs are
identified, conducts equity constraint and weighting analyses to provide further information about
the nature of those trade-offs. ECEA is particularly useful when health policy decisions have
substantial impacts on household financial risk protection, as is often the case in low- and middleincome countries. DCEA, by contrast, is better suited to cases where impacts on household income
and financial risk protection are less important, as is the case in high-income countries like the UK
with well-developed social protection safety nets and universal publicly funded health care systems.
Key messages for decision makers who use cost-effectiveness studies



Some decisions involve trade-offs between improving health and reducing health
inequity.
Such trade-offs can occur, for example, when delivering services to disadvantaged
communities requires additional cost, due to poor infrastructure or utilisation barriers.
In such cases, equity-informative CEA can help make fairer decisions.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis v
Key messages for analysts who produce cost-effectiveness studies




Who gains most depends on social variation in several factors including health risks,
access to care, adherence to care, quality of care, clinical effects, and capacity to
benefit.
Providing a full picture of health equity impacts requires analysing not only who gains
but also who bears the opportunity costs of diverting scarce resources from other uses.
Understanding health equity impacts can benefit from analysis of social inequality in
service delivery and the implementation costs of attempting to reduce this inequality.
Health equity findings should include sensitivity analysis that helps decision makers
understand the implications of alternative value judgements about equity.
vi CHE Research Paper 135
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 1
Introduction
Health equity has risen to prominence on policy agendas, in the wake of the universal health
coverage movement (Gwatkin and Ergo 2011, Norheim, Ottersen et al. 2014, Cotlear, Nagpal et al.
2015, Wagstaff, Cotlear et al. 2016) and landmark international reports on inequality in health
(Marmot, Friel et al. 2008, Marmot, Allen et al. 2012) and health care (WHO 2014, World Health
Organization 2015). However, the cost-effectiveness studies that are routinely used across the globe
to inform decision making in health care and public health rarely provide information about who
gains and who loses from alternative decisions (Sassi, Archard et al. 2001, Cookson, Drummond et al.
2009, Culyer 2012, Johri and Norheim 2012, Daniels, Porteny et al. 2016).
Equally, most studies of health equity lack a cost-effectiveness perspective. A cost-effectiveness
perspective can enhance health equity research in two important ways. First, as well as
understanding the nature and causes of health inequalities, decision makers need to know how the
policy options open to them are likely to impact upon health inequalities. This requires not only
epidemiological analysis of the determinants of health inequalities, but also economic analysis of
costs and how policies are likely to change individual and organisational behaviour. Second, when
evaluating the health equity impacts of policies the distribution of opportunity costs matters as well
as the distribution of benefits. If costs fall on government budgets for health, education or other
public services, for example, this is likely to have a disproportionate impact on disadvantaged
populations who rely most heavily on those public services.
This report brings together two different strands of literature – on cost-effectiveness analysis and on
health equity – and shows how they can be integrated in a practical way to provide decision makers
with useful information about health equity concerns. It describes methods for using costeffectiveness analysis to provide new information about inequalities in health-related outcomes that
may be considered unfair, or as ‘health inequities’.
In recent years there have been a number of methodological advances in this area which have been
developed into practical tools (Verguet, Murphy et al. 2013, Asaria, Griffin et al. 2015, Verguet,
Laxminarayan et al. 2015, Verguet, Kim et al. 2016). The technical details of these new methods
have been published in scientific journals in a form suitable for health economists, but may require
further description to be communicated in a more user-friendly manner suitable for the policy
makers and managers who commission and use cost-effectiveness evidence and for students and
scholars in disciplines outside health economics.
This report focuses on methods of analysis, rather than processes of decision making. Evidence
about health inequities generated using the methods described in this report is only one input into
decision making, and there are further important issues – not addressed in this report – about how
to design processes of decision making that appropriately reflect concerns about health inequities
(Culyer and Lomas 2006). Re-aligning the methods of CEA to address equity concerns is only one
facet of the much larger question of how to design fair processes of decision making that
appropriately address equity concerns (Culyer 2012, Culyer 2016). Equity-informative CEA can only
address an important subset of the diverse stakeholder concerns about fairness that may arise in
relation to a specific decision, and decision makers will always need to consider wider issues and
wider sources of information. One useful approach to ensuring that decision makers give due
attention to wider equity concerns, for example, is the use of ‘equity checklists’ (Culyer and
Bombard 2012, Norheim, Baltussen et al. 2014). More fundamentally, robust institutional
structures, processes and incentives are needed to ensure that decision makers take appropriate
steps to reduce inequities in health, as well as new methods of analysis. Analysis of the health
equity implications of decisions cannot help to improve decision making, for example, if it is poorly
2 CHE Research Paper 135
conducted or communicated, if it is based upon the idiosyncratic value judgements of a narrow
group of experts rather than a broader community of stakeholders, if policy advisers lack sufficient
training to understand the findings, or if the conclusions are disregarded by decision makers who
merely pay lip-service to health equity concerns.
Implicitly or explicitly, all CEA studies already embody social value judgements about equity in
scoping and methodological decisions about the relevant policy options and comparators, about
which costs and effects to measure, about how to compare costs and effects of different kinds,
about how to aggregate costs and effects for different people and organisations, about how to value
future costs and effects, and so on (Shah, Cookson et al. 2013). To take just one example, a
methodological decision to include ‘productivity costs’ in the cost denominator will increase the
implicit priority given to life-extending treatments for ‘productive’ citizens, such as high-earning
parents of young children, compared with ‘unproductive’ citizens, such as frail older people. Indeed,
the basic ethical underpinning of standard cost-effectiveness analysis adheres to an equity principle
– the principle that health policy makers should seek to increase sum total population health as
much as possible given scarce resources, and hence that each unit of health gain should be valued
equally, regardless of who obtains it (Cookson 2015). These value judgements are rarely mentioned
in applied CEA studies or health technology assessment (HTA) reports, but are extensively discussed
in textbooks, methods guidance documents and other underpinning literature (Culyer 2016). This
article shows how to go beyond embodying pre-specified value judgements about equity within
applied CEA studies, and towards using the techniques of CEA to provide new information about the
health equity implications of alternative policy options that facilitate deliberation among decision
makers and stakeholders (Culyer 2012, Culyer 2016).
This report has two main sections. In the first section, we introduce key concepts of costeffectiveness analysis and health equity. In the second section we then describe three general
approaches and two specific methods for using cost-effectiveness analysis to address health equity
concerns. These approaches are illustrated with applications to low-, middle- and high-income
country settings.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 3
Part I: Concepts
Cost-effectiveness analysis
Cost-effectiveness analysis compares the costs and effects of one policy option with another policy
option – which might be ‘do nothing’ (Drummond, Sculpher et al. 2015). Health effects are often
measured using a composite summary index of health, such as the quality-adjusted life year (QALY)
or the disability-adjusted life year (DALY), to facilitate comparison between policies in different
disease areas with diverse and distinct mortality and morbidity impacts. This allows the calculation
of a cost per QALY gained, or cost per DALY averted. A cost-increasing policy option can then be
considered ‘cost-effective’ if its cost per unit of health gain compares favourably with alternative
ways of using scarce resources.
In a public health system with a fixed budget, the displaced activities will comprise alternative health
programs that would have produced alternative health benefits. Cost-effectiveness can then be
defined as a test of whether a program will improve total health. A cost-effective policy will have a
positive net health impact, because its health gains will outweigh the health losses from shifting
expenditure away from other health programs. By contrast, a cost-ineffective policy will have a
negative net health impact, because the health losses from shifting expenditure away from other
health programs will outweigh the health gains. The policy objective underpinning conventional CEA
can then be interpreted as the quasi-utilitarian health equity objective of maximising sum total
health in the general population (Culyer 2006, Cookson 2015). CEA can thus help decision makers to
choose investments that increase total health, and avoid investments that reduce total health. This
interpretation of opportunity costs in terms of forgone health benefits is more problematic if there is
no fixed health budget, in which case opportunity costs may instead fall on household consumption
(via increased taxes or insurance premiums) or on reductions in public expenditure on programs not
primarily designed to improve health. Whatever the setting in which CEA is used, however, the
recognition of opportunity costs – i.e. that resources used in the provision of a program would have
generated value if used elsewhere – is fundamental to CEA. Every benefit attributed to a program
must be weighed against those displaced when resources are diverted from alternative activities.
To provide useful information for priority setting decisions, it is important to give a clear description
of the options being compared and the policy context. Decision makers can then assess how far
these options are relevant to their own decision making context. Comparisons of hypothetical policy
options may have limited use in helping with the task of setting priorities between specific health
sector investment decisions. However, they can help with the task of ‘agenda setting’ or ‘making the
case’ in favour of investment in a broad policy area (e.g. mental health) or a broad objective or
direction of travel for policy making. For example, a 2013 report by the Lancet Commission on
Investing in Health made the case that low- and middle- income countries should all set a long-term
policy ambition of achieving universal health coverage by the year 2035 (Jamison, Summers et al.
2013). This case was supported by an economic evaluation comparing two hypothetical policy
options – no growth in health spending versus sustained long-term growth so as to achieve universal
health coverage by 2035.
Cost-effectiveness studies of specific policy alternatives can help to inform priority setting in at least
four decision contexts, each of which involves different though overlapping communities of decision
makers, analysts and stakeholders. Decisions that are considered investments should also consider
disinvestments as these are likely to have distributional consequences as well:
1.
Pricing and reimbursement of new pharmaceuticals and other health technologies – e.g.
whether to fund imatinib for stomach cancer, at what price, and for which patients?
4 CHE Research Paper 135
2.
3.
4.
Health care benefit package design – e.g. whether to cover diabetes in the national
health insurance plan, and if so which diabetes interventions to include in the package?
Investments in healthcare organisation and delivery infrastructure – e.g. whether to
invest in a community health worker programme, and if so how many and what kinds of
new staff to recruit, located where, and with what pay and conditions?
Public health investments and regulations – e.g. whether to implement a sugar tax, and
if so how much should it be?
In each case, cost-effectiveness analysis can provide decision makers with a standard way of
assessing the value for money of specific policy options compared with alternative uses of scarce
resources. In (1), for example, the funding for imatinib could be used for other hospital-based
cancer treatments – or, with more difficulty, shifted towards primary care. In (2), there may only be
enough money to cover diabetes or dementia – but not both. In (3), the investment could instead
be used to buy hospital equipment. And in (4), the opportunity costs of a sugar tax will be felt by
consumers paying higher food prices – potentially pushing some further into poverty. In each case,
equity plays an important but typically unanalysed role: standard cost-effectiveness analysis does
not tell us who gains most, nor who bears the heaviest opportunity costs.
Health equity
In this report, we use the term ‘health inequities’ in a broad sense to refer to differences in healthrelated outcomes that may be considered unfair, inequitable or unjust (Whitehead 1992, Norheim,
Baltussen et al. 2014). Health equity then refers to the policy goal of seeking to reduce or eliminate
health inequities. Under our broad definition, health inequities may include not only inequalities in
health per se but also inequalities in outcomes related to ill-health and lack of access to affordable
health care, such as catastrophic health care expenditure, impoverishment, inability to work, and
inability to perform domestic tasks such as child care. These outcomes may be unequally distributed
– for example, decisions about public funding for a particular medical technology could have
differential impacts by social group on healthcare costs, or on ability to work or perform domestic
tasks, and the resulting financial risks to households.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 5
Figure 1 shows differences in people’s lifetime experience of health in England and Ethiopia in the
year 2011 (Dawkins 2015, Asaria et al. 2015, Love-Koh. 2015); see Annex.
90
Healthy life Expectancy
80
70
64.7
68.6
73.6
75.6
60.3
60
50
70.6
47.5
51.0
53.7
55.6
England
40
Ethiopia
30
20
10
0
Q1
Q2
Q3
Q4
Health Quintile Group
Q5
Figure 1: Distribution of Healthy Life Expectancy in England and Ethiopia in 2011
Note: This figure compares quintile groups of healthy life expectancy at birth as predicted by socioeconomic status and sex
only.
The figure shows healthy life expectancy at birth by health quintile groups. The health quintile
groups are based on healthy life expectancy at birth as predicted by just two equity-relevant
variables: socioeconomic status and sex. We use health quintile groups, rather than socioeconomic
groups, to emphasise that policy makers may be concerned with many different sources of
inequality in health. In principle, we could also include variation due to further equity-relevant
variables such as ethnicity, region, working conditions and so on. The figure shows that there is a
gradient in health within each country, but also that there is an inequality between countries since
the most advantaged health group in Ethiopia does not achieve as much health as the least
advantaged in England. If considered unfair, the differences in Figure 1 represent health inequities.
Differences in health-related outcomes are sometimes considered fair, and equality in health-related
outcomes are sometimes considered unfair. For example, providing different health services to
people with different needs may be fair, and providing the same services to people with different
needs may be unfair. More controversially, if someone’s life is cut short by a random accident – a
matter of ‘residual’ or ‘unexplained’ bad luck, not avoidable or remediable by any social action – this
might by some be considered unfortunate or tragic, but not a matter of social injustice (Hausman
2013). This can make a big difference to inequality analysis. For example, if we consider
‘unexplained’ variation to be unfair, we would need to include it in Figure 1 and predict the ‘ex post’
distribution of healthy length of life, allowing for unexplained variation in mortality and morbidity,
which would include much more variation than the ‘ex ante’ distribution of healthy life expectancy
(Asada, Hurley et al. 2015).
6 CHE Research Paper 135
In assessing how far a particular difference is ‘unfair’, there are always thorny issues of both
scientific and social value judgement to consider. For example, inequality in life expectancy may be
considered more ‘unfair’ when it is caused by unequal childhood circumstances than by lifestyle
choices in adulthood. Judgements of this kind turn on contestable scientific judgements about
complex causal pathways in the face of imperfect evidence, as well as contestable ethical issues of
social value judgement. That is why we use the cautious phrase ‘differences that may be considered
unfair’ in our definition of inequity, rather than the bolder phrase ‘unfair differences’ – to emphasise
that value judgements about equity are properly a matter for decision makers and stakeholders
rather than analysts.
We also use the less well-known term, ‘health equity impacts’, to mean changes in health inequities
brought about by policy interventions. In practice, due to data limitations, modelling of health
equity impacts is typically restricted to differences associated with categorical social group variables
such as income or wealth groups, education groups, ethnic groups, by gender or region. For
example, a social variable commonly used in low- and middle-income country applications is wealth
quintile based on a survey of household assets, and a social variable commonly used in high income
country applications is socioeconomic quintile group based on small area deprivation. These were
the socioeconomic variables used to estimate the figures in Figure 1 for Ethiopia and England,
respectively. When suitable data are available, health equity impacts can also include impacts on
within-group differences associated with continuous social variables, such as income, or even
impacts on the whole-population univariate distribution of health unrelated to any particular social
variable.
There are many reasons for policy concern about health inequalities, one of which is that all forms of
social inequality – including health inequality – may act as a kind of ‘social pollution’ that is bad for
everyone’s health in society (Wilkinson and Pickett 2010). In principle, if a particular decision had
large impacts on the distribution of wealth and power within society, it might be worth
incorporating the resulting change in the level of ‘social pollution’ into cost-effectiveness analysis.
When evaluating most health policies, however, it is reasonable to assume that impacts on social
inequality as a whole will be relatively small and hence that any indirect impacts on health via
changes in the level of ‘social pollution’ will be too small to be worth including in the analysis.
The tools described in this report can be used to provide information about health equity impacts for
decision makers and stakeholders; it is then up to them to process the information, draw
conclusions and take decisions based on their own value judgements.
Accounting for the social distribution of opportunity costs
The distribution of policy costs matters, as well as the distribution of policy benefits. Understanding
health equity impacts requires analysing not only who gains health benefits but also who bears the
opportunity costs – the forgone health benefits that could have been generated through alternative
ways of using the same scarce resources may also be unequally distributed. The distribution of
opportunity costs will depend crucially on where the funding for the policy or programme comes
from. For example, if the money comes from an increase in progressive general taxation, the
absolute opportunity costs are likely to be borne disproportionately by the rich and the opportunity
costs in terms of losses in health and wellbeing may be equally distributed. By contrast, if it is
funded by reducing public expenditure on other health, education or welfare services, the
opportunity costs in terms of losses in health and wellbeing may be borne disproportionately by
poorer individuals who rely more heavily upon public services. The same applies to foreign aid
funding that would otherwise be used to fund alternative programs that disproportionately benefit
more socially disadvantaged people.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 7
In order to estimate net health equity impact accurately one needs to compare the health benefits
and health opportunity costs using the same metric. To do this, it is usually necessary to measure
health outcomes using a generic health measure that can be compared across different diseases,
such as QALYs or DALYs, rather than in disease specific units such as cases of malaria prevented.
This is because we typically do not know which patient groups will bear the health opportunity costs,
and without a generic measure of health it is hard to compare the value of a case of malaria
prevented versus a case of cancer or tuberculosis or heart disease. The only case in which it might
make sense to compute health benefits and opportunity costs in the same disease specific unit is if
there is a fixed budget for a ‘vertical programme’ in a particular disease area, which cannot be used
for treating any other disease. In that case, the opportunity costs of alternative uses of the budget
will all fall on patients with that specific disease, and so it may make sense to measure both health
benefits and opportunity costs in disease specific units. Where reliable data on morbidity impacts
are not available for computing QALYs or DALYs, an alternative is to measure health benefits in
terms of life years or mortality risk reduction. However, this is less comparable with health
opportunity costs because some interventions focus primarily on reducing morbidity rather than
mortality, and the balance between the two can vary substantially between different disease areas.
Insofar as public and donor expenditure tends disproportionately to benefit socially disadvantaged
groups, this provides a tough benchmark for assessing the health equity impacts of specific public or
donor-funded programmes compared with other uses of public or donor funding. As Figure 2
shows, interventions that may initially seem to have a ‘pro-poor’ health equity impact may in fact be
equity-neutral or even ‘anti-poor’, when one considers where the money is coming from and what
the alternative uses of that money would be. This is shown with the downward arrows that convert
the ‘gross’ health impacts – considering programme benefits only – to ‘net’ health impact, after
allowing for the health opportunity costs of alternative uses of the money.
8 CHE Research Paper 135
Case 1: Equal Opportunity Costs
40,000
Health Benefit (e.g. DALYs)
Gross Health Benefit
30,000
20,000
10,000
0
Net Health Benefit
-10,000
-20,000
0
1
Poorest
2
3
4
Wealth
WealthQuintiles
Quintiles
5
Richest
Case 2: Unequal Opportunity Costs (Larger for the Poor)
Health Benefit (e.g. DALYs)
40,000
35,000
Gross Health Benefit
30,000
25,000
20,000
15,000
10,000
Net Health Benefit
5,000
0
-5,000
0
1
Poorest
2
3
Wealth
WealthQuintiles
Quintiles
4
5
Richest
Figure 2. Distribution of Net Health Benefits from a Hypothetical Programme
The top panel of Figure 2 illustrates the assumption that health opportunity costs (forgone health
benefits) are constant across all groups. For an investment in a hypothetical programme the gross
health benefits as seen in the top lines of each case in Figure 2, show the same distribution of
benefit. In case 1, as the opportunity costs are accounted for by taking the forgone benefits to be
constant across wealth quintiles leads one to conclude there is a ‘pro-poor’ distribution of net health
benefits. That is, the poor appear to benefit more in the net benefits of case 1 with negative net
health benefits for the richest. The bottom panel illustrates the assumption that the opportunity
costs are unequally distributed and disproportionately borne by more socially disadvantaged groups.
The result is now a ‘pro-rich’ distribution of net health benefits, with negative net health benefits for
the poorest. This public programme is less pro-poor than alternative public programmes. So if the
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 9
goal is to reduce health inequality, the money would be better spent on other public programmes
rather than this one.
Empirically estimating the distribution of opportunity costs is difficult. However, work is under way
by researchers at the University of York to estimate the social distribution of the opportunity costs of
public healthcare expenditure in England, and similar studies in the future could provide other
relevant estimates to be incorporated in future work.
Trade-offs between total health and health equity
Health maximisation and health equity are often aligned, but not always. The ‘health equity impact
plane’ in Figure 3 helps one to think about the potential trade-offs (McAuley, Denny et al. 2016). The
vertical axis shows the cost-effectiveness of a health intervention. As explained above, this can
under certain assumptions be interpreted as its ‘net total health impact’ – i.e. the total health
benefits of the programme minus the forgone health benefits that would have been obtained by
spending the money on other health programmes. The horizontal axis shows the ‘net health equity
impact’, which refers to the net impact upon fairness in the distribution of health after allowing for
opportunity costs as well as benefits. Health equity impact can be defined and measured in various
ways, as discussed below.
Cost-Effectiveness
(Net Total Health Impact)
+
II. Win-Lose
I. Win-Win
Cost-effective
Harms equity
Cost-effective
Improves equity
ve
+
III. Lose-Lose
IV. Lose-Win
Cost-ineffective
Harms equity
Cost-ineffective
Improves equity
Net
Health
Equity
Impact
Figure 3: Health Equity Impact Plane
The plane defines four quadrants. In quadrant I (‘win-win’), the policy improves both total health
and equity, and in quadrant III (‘lose-lose’), the policy harms both. In these two cases, health
maximisation and equity are aligned. In the other two quadrants, by contrast, health maximisation
and equity are opposed and there may be trade-offs. In quadrant II (‘win-lose’), the policy is good
for total health but bad for equity, and in quadrant IV (‘lose-win’), the policy is bad for total health
but good for equity.
10 CHE Research Paper 135
If all policies fell in the two aligned quadrants (win-win and lose-lose) then there would be no need
to analyse health equity impacts. We would then have a guarantee that a policy identified as costeffective using standard CEA approaches would always improve health equity, and a cost-ineffective
policy would always harm health equity.
Many policies do indeed fall into these two quadrants. In low-income countries, for example,
investments in high-cost hospital treatments often fall into the ‘lose-lose’ quadrant of being neither
cost-effective nor likely to reduce health inequality, insofar as they deliver relatively small health
gains per unit of cost and disproportionately benefit wealthy urban elites. By contrast, programmes
of vaccination and primary care for infectious diseases often fall into the ‘win-win’ quadrant of
delivering large health gains per unit cost and reducing health inequality, insofar as they
disproportionately benefit socially disadvantaged groups at high risk of infection.
However, socially disadvantaged groups may sometimes gain less than more advantaged groups
from a decision to fund a particular medical technology, due to unequal access, utilisation,
adherence, quality and outcomes of healthcare (Tugwell, de Savigny et al. 2006). For example,
access costs may be relatively high, and effective healthcare coverage relatively low, in urban slums
and remote rural areas which lack well-resourced clinics and struggle to recruit qualified medical
staff. In such cases, there may be trade-offs between health maximisation and health equity – and
hard choices between more equal delivery versus larger total health gains.
Measuring health equity impacts
Health inequities can be measured in different ways depending on the decision problem in hand and
the equity concerns of the relevant decision makers and stakeholders. There are many useful
training resources describing how to do so, in general (O'Donnell, van Doorslaer et al. 2008,
Fleurbaey and Schokkaert 2011, Asada, Hurley et al. 2014), in the context of health system
performance monitoring (World Health Organization 2013) and in the context of cost-effectiveness
analysis (Asaria, Griffin et al. 2013). In each case, at least four basic questions must be addressed:
1.
2.
3.
4.
Equality of what?
Equality between whom?
Equality indexed how?
Equality adjusted how?
In relation to ‘equality of what?’, for example, the central equity concern might be inequality in
health, or inequality in capacity to benefit, or inequality in financial risk protection; and there are
many different metrics for measuring each of these concepts. In relation to ‘equality between
whom?’, for example, the unit of analysis might be individuals or households or geographical areas
or social groups, and the inequality breakdowns of interest may relate to socioeconomic status,
ethnicity, gender or other social variables. In relation to ‘equality indexed how?’, different indices of
inequality may contain different normative judgments and can yield quite different patterns of
change depending for example whether they are defined in terms of relative or absolute differences
and in terms of achievement or shortfall (e.g. survival or mortality) (Harper, King et al. 2010) (Arcaya,
Arcaya et al. 2015, Kjellsson, Gerdtham et al. 2015, Wagstaff 2015). In relation to ‘equality adjusted
how’, it is rarely enough to measure crude, unadjusted differences in health-related outcomes: one
should also consider adjusting for factors that influence the assessment of how far differences are
unfair (Asada, Hurley et al. 2014, Asada, Hurley et al. 2015). For example, differences in health care
utilisation may be fair if they are due to difference in needs or preferences; and differences in health
outcomes may be fair if they are due to differences in exogenous factors beyond the control of
health services, such as a patient’s age. This adjustment process can yield different answers
depending on the selection of standardising variables, the choice of reference values for those
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 11
variables, and different model specifications and methods of adjustment (e.g. direct versus indirect
standardisation). To measure health inequality that is considered unfair, for example, differences in
mortality risk between socioeconomic groups may be partly driven by (or masked by) differences in
the age structure of the different groups. Insofar as age is not considered to be an ‘unfair’
determinant of mortality, an adjustment is needed to control for the influence of age on mortality.
There are various more or less sophisticated ways of adjusting for ‘fair’ differences in outcomes, but
whatever approach is taken it is important to be as clear and explicit as possible about the
contestable value judgements about fairness that underlie such adjustments.
Health equity concerns can be placed on a spectrum from general to specific. At the broad end of
this spectrum is concern to reduce unfair inequality in lifetime health between all individuals
(Norheim 2010, Robberstad and Norheim 2011). At the specific end of the spectrum might be
concern for equalising the benefits of a particular policy between urban and rural groups over a
particular time period. There are at least four aspects of health equity concern that can vary
depending on whether a more general or specific perspective is adopted:
1.
2.
3.
4.
Levels vs. gains – concern for equality in health levels (more general) rather than
concern for the health gains arising from a specific policy (more specific)
Outcome specificity – concern for equality in health (more general) rather than concern
for equality in the ill-health burden of cancer (more specific)
Time horizon – concern for equality over the life course (more general) rather than
equality in a particular age group or over a specific time period (more specific)
Single vs. multiple social variables – concern for inequality by multiple social variables
(more general) rather than by one social variable (e.g. urban vs. rural)
12 CHE Research Paper 135
Figure 4 illustrates a specific equity impact, using a modified version of a hypothetical example used
in a 2004 article in the Lancet (Gwatkin, Bhuiya et al. 2004). It shows two policies for increasing
coverage of skilled birth attendance in a low-income country – a ‘universal’ policy and a ‘targeted’
policy that focuses on developing maternity service infrastructure in socially disadvantaged areas. It
shows that the ‘targeted’ policy is less equal in terms of the distribution of gains: it delivers larger
benefits to socially disadvantaged groups, in terms of increases in the proportion of birth deliveries
attended by a skilled midwife. From this highly specific perspective, the ‘targeted’ policy may seem
unfair to people in the middle and upper wealth groups who gain less.
Gain in % attended deliveries
0.45
0.4
0.35
0.3
0.25
0.2
Standard
0.15
Targeted
0.1
0.05
0
Q1
(poorest)
Q2
Q3
Q4
Q5
Average
(richest)
Wealth Group
Figure 4: Coverage gains by wealth group, comparing standard and targeted options for improving maternity
services in a low income country – hypothetical example based on (Gwatkin, Bhuiya et al. 2004)
However, Figure 5 shows that the ‘targeted’ policy is more equal in terms of the resulting levels of
attended deliveries. If one takes this broader perspective, the ‘targeted’ policy may seem fairer.
1
% Attended Deliveries
0.9
0.8
0.7
0.6
0.5
Baseline
0.4
Standard
0.3
Targeted
0.2
0.1
0
Q1 (poorest)
Q2
Q3
Q4
Q5 (richest)
Wealth Group
Figure 5: Coverage levels by wealth group, comparing standard and targeted options for improving
maternity services in a low-income country – hypothetical example
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 13
One might also wish to take an even broader perspective, by looking at impacts on inequalities in
lifetime health of the kind illustrated by Figure 1 earlier in this report. From this perspective, the
‘targeted’ policy will seem even fairer, since it will help to reduce the large pre-existing inequality in
lifetime health between socioeconomic groups. This is easy to show using mathematical indices of
inequality. However, an important challenge for communicating lifetime health impacts to decision
makers is that individual policy decisions typically have small impacts on expected lifetime outcomes
for the average citizen. Health equity impacts still matter, however, even though they seem small
from the ‘ex ante’ perspective of expected changes in risk of future mortality and morbidity – since
we are still talking about large and important changes in many people’s lives from an ‘ex post’
perspective looking back at how far they experienced a long and healthy life (Eyal, Hurst et al. 2013).
From an ‘ex ante’ population perspective, the cost-effectiveness impacts of most policy decisions
also seem small – costing little to the average citizen and adding only a small proportion to average
healthy life expectancy. Furthermore, a sustained sequence of small impacts can over time
accumulate into a large impact.
Quantifying health equity trade-offs
When there are trade-offs between health equity and total health – i.e. we know that the policy lies
in either the ‘win-lose’ or ‘lose-win’ quadrants of Figure 3 – it can be helpful to provide a more
refined quantitative analysis of those trade-offs. One general approach, which we call ‘equity
constraint analysis’, is to count the cost of choosing a fairer but less cost-effective policy, in terms of
a reduced total health (QALY equality) benefit. Another general approach, which we call ‘equity
weighting analysis’, is to quantify how much concern for health inequity is required to recommend a
fairer but less cost-effective policy option, using sensitivity analysis. This sensitivity analysis can be
done by specifying alternative ‘equity weights’ for health benefits to different people, or alternative
values of an ‘inequality aversion parameter’ quantifying your degree of concern for reducing health
inequality versus improving total health (Norheim 2013) (Asaria, Griffin et al. 2013, Wagstaff, Cotlear
et al. 2016). Sensitivity analysis can help decision makers understand the implications of alternative
value judgements. The aim of health equity trade-off analysis is thus to help inform a decision
making process, rather than to impose a particular set of value judgements upon decision makers.
14 CHE Research Paper 135
Part II: Methods
This part of the report describes three general ways of using cost-effectiveness analysis to evaluate
policy concerns about health equity, each of which addresses different equity questions (Cookson,
Drummond et al. 2009, Johri and Norheim 2012). These are: equity impact analysis, equity
constraint analysis and equity trade-off analysis. Table 1 provides a definition of each approach and
the questions they help to answer.
Table 1. Three General Approaches to Evaluating Health Equity Concerns
Approach
Equity impact
analysis
Description
Disaggregation of the relevant
costs and benefits by equityrelevant sub-groups, providing
a dashboard of results.
2.
Equity constraint
analysis
3.
Equity weighting
analysis
Calculation of the health
opportunity cost of choosing a
fairer option rather than a
more cost-effective option.
Sensitivity analysis around the
value of health equity impacts,
based on different concepts of
inequality, and the strength of
concern for reducing health
inequality.
1.
Questions Answered
How much do different social groups gain or lose?
This may be in terms of money, health services, health
outcomes or other outcomes related to ill-health and
access to affordable health services, such as financial
risk protection.
How much total health benefit is forgone if a more
cost-effective option is ruled out on equity grounds?
How large is the health equity impact in terms of
standard summary metrics of inequality? How much
concern for health equity is required to choose a more
equitable option compared with a more cost-effective
option?
We use the umbrella term ‘equity-informative cost-effectiveness analysis’ to describe any study that
uses the methods of cost-effectiveness analysis to provides information about health equity
objectives other than improving total health. We illustrate two specific methods of equityinformative cost-effectiveness analysis. Extended Cost-Effectiveness Analysis (ECEA), developed by
the Disease Control Priorities, 3rd edition project (www.dcp-3.org), examines the social distribution
of costs, health effects and financial risk protection effects. Distributional Cost-Effectiveness Analysis
(DCEA), developed by the University of York, examines the distribution of health benefits and
opportunity costs and then, if health equity trade-offs are identified, conducts equity constraint and
equity trade-off analyses to provide further information about the nature of those trade-offs. The
key difference between these two methods is that ECEA presents estimation of financial risk
protection (e.g. poverty reduction benefits) and disaggregated information on a range of different
costs and effects, while DCEA aggregates and compares all costs and effects using the common
currency of QALYs or DALYs, thus allowing the construction of summary indices of inequaity and
equity weighting analysis.
We first describe methods of equity evidence review that can be performed alongside or prior to
cost-effectiveness analysis in order to provide information about health equity concerns.
Equity evidence review
Equity evidence review can address a range of health equity questions of interest to decision
makers, for example ‘What are the relevant health equity issues?’, ‘What do stakeholders think?’,
‘What is already known about the size of pre-existing health inequities?’ and ‘What do studies of
past policies suggest may be the potential health equity impacts of the policy options currently
under consideration?’.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 15
There are published guidelines for using the methods of systematic review of randomised controlled
trials to gather information about equity impacts, such as the PROGRESS framework adopted by
both the Campbell and Cochrane Collaborations (O'Neill, Tabish et al. 2014). Box 1 describes a
Cochrane systematic review regarding food supplementation for disadvantaged young children;
another example is the review by Brown and colleagues of the equity impact of interventions and
policies to reduce smoking in youth (Brown, Platt et al. 2014) and another is the review by Noor and
colleagues about reducing inequity in use of insecticide-treated bed nets in Kenya (Noor, Amin et al.
2007). Systematic review methods have an important role in helping to avoid the all-too-common
trap of falling unknowingly into a selective reading of the evidence, and to avoid ‘re-inventing the
wheel’. However, systematic reviews can be resource intensive, and often require several months of
work by large multi-disciplinary teams. Furthermore, there are ongoing debates about appropriate
review methods for complex interventions where theoretical and observational evidence may play a
more central role than randomised controlled trials. So if the requisite capacity is not available to
conduct a systematic review, it may be useful to undertake a rapid review.
It is also possible to conduct a review of philosophical arguments about equity in relation to a
particular case, such as the ‘ethical case review’ approach that Miljeteig and colleagues used to
examine end-of-life decisions in a neonatal unit in India (Miljeteig, Johansson et al. 2010). However,
in conducting a review of this kind analysts must be wary of going too far down the road of
adjudicating between alternative views of justice in an attempt to answer the question ‘what is
fair?’, rather than clarifying alternative points of view so that decision makers can make up their own
minds.
Box
anan
equity
evidence
review
Box 11––Example
Exampleofof
equity
evidence
review
Foodsupplementation
supplementation
disadvantaged
young
children
– a systematic
Food
forfor
disadvantaged
young
children
– a systematic
review review
Kristjanssonand
andcolleagues
colleagues
Cochrane
systematic
method
to examine
Kristjansson
useuse
thethe
Cochrane
systematic
reviewreview
method
to examine
interventions to
interventions
to address
malnutrition
in young
children.
primarily review
the effectiveness
address
malnutrition
in young
children. They
primarily
review They
the effectiveness
of supplementary
feeding
interventions,
but they
secondarily look
howsecondarily
such interventions
withinterventions
inequalities,
of supplementary
feeding
interventions,
butatthey
look atinteract
how such
implementation
and
adverse
outcomes.
We
focus
in
this
synopsis
on
what
they
find
for
inequality.
interact with inequalities, implementation and adverse outcomes. We focus in this
synopsis on
what they find for inequality.
The authors develop a conceptual framework to guide thinking about how inequalities may factor into
childhood
nutrition
andahealth.
Equityframework
was then evaluated
looking about
at outcomes
for given subgroups
The authors
develop
conceptual
to guidebythinking
how inequalities
may
such as age, sex, level of malnourishment (baseline health) and socioeconomic status.
factor into childhood nutrition and health. Equity was then evaluated by looking at outcomes
for given subgroups
such as
age,
sex, level
of malnourishment
(baseline
health) and
Meta-analysis
of the studies
and
differences
between
sub-groups were
then conducted.
The results of
socioeconomic
status.find that supplementary feeding programmes have unequal impacts on certain
this
review and analysis
subgroups, being more effective for children who were poorer and more malnourished, but no difference
Meta-analysis
of subgroups
the studies
andasdifferences
of
impact in other
such
by sex of thebetween
child. sub-groups were then conducted. The
results of this review and analysis find that supplementary feeding programmes have unequal
Reference:
E, Francis being
DK, Liberato
Benkhalti
M, Welch
Batalpoorer
M, Greenhalgh
T,
impacts onKristjansson
certain subgroups,
more S,
effective
forJandu
children
who V,
were
and more
Rader
T,
Noonan
E,
Shea
B,
Janzen
L,
Wells
GA,
Petticrew
M.
Food
supplementation
for
improving
the
malnourished, but no difference of impact in other subgroups such as by sex of the child.
physical and psychosocial health of socio-economically disadvantaged children aged three months to five
years.
Cochrane
Database of
2015,
Issue. DOI:Jandu
10.1002/14651858.CD009924.pub2.
Reference:
Kristjansson
E, Systematic
Francis DK,Reviews
Liberato
S, Benkhalti
M, Welch V, Batal M,
16 CHE Research Paper 135
Equity impact analysis
Equity impact analysis examines who gains and who loses from one policy option compared with
another, and by how much. Typically, this approach yields a ‘dashboard’ of outcomes, showing gains
and losses by social group for various different kinds of costs and effects. This provides the decisionmaker with relevant disaggregated information on the equity impacts.
An example of equity impact analysis alongside cost-effectiveness analysis in a high income context
is the study by Holmes and colleagues of the impacts of minimum alcohol pricing in the UK (Holmes,
Meng et al. 2014). This study estimated the social class distribution of effects on alcohol
consumption, spending and alcohol-related health harm, and found that benefits are substantially
concentrated on heavy drinkers in routine and manual worker households.
The DCP3 project has developed a specific method of cost-effectiveness analysis that incorporates
equity impact analysis and analysis of financial risk protection effects as well as health effects, called
‘extended cost-effectiveness analysis’ (ECEA) (Verguet, Laxminarayan et al. 2015). This method has
now been applied to about 20 studies of policy interventions in several different low- and middleincome country settings, producing breakdowns of costs, health benefits and financial risk
protection benefits by socioeconomic quintile group (Verguet and Jamison forthcoming). Box 2
gives an example of ECEA relating to tobacco taxes in China.
Sometimes equity impact analysis only examines the benefits of a policy, not the costs. As explained
above, this provides an incomplete picture of health equity impact, since it only looks at one side of
the distributional coin: the costs may also be unequally distributed.
Equity impact analyses can also be performed outside the context of cost-effectiveness analysis, to
look at the impacts of changes in risk factors, health behaviours or the utilisation of effective health
technologies, rather than the impacts of specific policy options. For example, Bajekal and colleagues
examined the impacts on coronary heart mortality in England from 2000 to 2007 of changes in risk
factors and treatment utilisation in different social groups (Bajekal, Scholes et al. 2012). This kind of
study does not directly inform the priority setting task of selecting between specific future policies,
as it does not provide information about either the costs of those policies or their effects on risk
factors, health behaviours or the utilisation of effective health technologies. However, it can provide
decision makers with useful contextual information for policy making purposes, in raising awareness
of the importance of particular factors and their contributions to health inequality.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 17
Box 2 – Example of an equity impact analysis
Tobacco tax in China – an “extended cost-effectiveness analysis” (ECEA)
This modelling exercise examines a 50% excise tax on cigarettes over a 50-year period. In aggregate, this is
estimated to save 231 million life years, add $703 billion in tax revenue to government budgets, reduce
expenditure on tobacco-related illness and improve financial risk protection.
The ECEA further breaks down model inputs by equity-relevant groups, in this case by five income quintiles. The
descriptive tables of the paper show how input variables such as smoking prevalence, cigarette consumption,
price elasticity and health utilization are all parametrized according to income quintile.
The results are presented in a dashboard-like format, where the aggregate results are broken down into results
for each quintile group. For example, the years of life gained are more concentrated on the poor (79 million in
the poorest quintile) than the rich (11 million in the richest quintile), as seen in the figure below. Expenditures
on tobacco, which one may worry would be regressive with a new excise tax, also show a progressive
distribution in absolute terms, or an inverted U distribution in relative terms. In either case, the tax is shown not
to overly harm the poor.
Lastly, the ECEA also incorporates financial risk protection benefits. In the case of the Chinese tobacco tax, this is
measured as a money-metric value of insurance. This is similarly disaggregated by income quintile, showing that
there is more financial risk protection for the poorer quintiles.
Reference: Verguet S, Gauvreau CL, Mishra S, MacLennan M, Murphy SM, Brouwer ED, Nugent RA, Zhao K, Jha
P, Jamison DT. The consequences of tobacco tax on household health and finances in rich and poor smokers in
China: an extended cost-effectiveness analysis. The Lancet Global Health. 2015 Apr 30;3(4):e206-16.
18 CHE Research Paper 135
Finally, a form of equity impact analysis known as ‘benefit incidence analysis’ looks at the benefits of
public healthcare spending as a whole for different social groups. Traditionally, this kind of analysis
has typically assumed that health benefits are proportional to health care consumption, and has only
looked at the average benefits of the current overall level of spending rather than the marginal
benefits of changes in spending. So traditional forms of benefit incidence analysis have not been
useful for predicting the health benefits of a policy decision to change expenditure from the current
baseline level. However, some benefit incidence analyses are now starting to look at marginal
benefits, by exploiting data on sub-national variation and change in expenditure and outcomes, and
so can be more useful for priority setting purposes (Kruse, Pradhan et al. 2012).
Equity constraint analysis
Equity can be analysed as a constraint on the pursuit of cost-effectiveness, rather than as a goal or
outcome to be pursued in its own right. The health opportunity cost of imposing this ethical
constraint can be calculated as the difference in total health benefit between the most cost-effective
policy option and a more equitable policy option. The health loss associated with choosing a more
equitable option thus gives an indication of the value the decision maker places on equity (Williams
and Cookson 2006). This approach can be implemented either using a simple cost-effectiveness
framework comparing two or more options given a fixed budget, or using more specialised
‘mathematical programming’ techniques to handle complex choices involving different amounts of
expenditure on different programmes (Earnshaw, Richter et al. 2002, Epstein, Chalabi et al. 2007).
Box 3 – Example of an equity constraint analysis
HIV Treatment in South Africa – a mathematical programming study
Cleary and colleague analyse the equity concern to provide only the most effective treatment as a
constraint on improving total health in relation to anti-retroviral (ART) HIV treatment in South Africa.
They measure cost-effectiveness in terms of QALYs gained per unit of cost, and equity in terms of the
percentage of met need for ART.
The study uses mathematical programming to examine different treatment delivery scenarios under
different health budgets. As budgets increase, more total health can be provided. However, at given
budget levels there are trade-offs and more cost-effective solutions exist that deliver larger total health
gains by offering some people cheaper but less effective treatments rather than fully meeting their needs
by proving the most effective treatment. Hence particular target levels of met need (the equity
constraint) are only achievable by sacrificing total health.
This approach highlights that the opportunity cost of equity can be understood in terms of the amount of
health that is forgone by implementing a more equitable solution, as opposed to implementing the most
cost-effective solution under a given budget constraint.
Reference: Cleary, S., Mooney, G., & McIntyre, D. (2010). Equity and efficiency in HIV-treatment in South
Africa: the contribution of mathematical programming to priority setting. Health Economics, 19(10),
1166-1180. doi:10.1002/hec.1542
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 19
Equity weighting analysis
The final approach is health equity weighting analysis. In contrast to the previous approaches, this
approach attempts to quantify the overall health equity impact, using summary indices of inequality,
and to analyse trade-offs between the health equity impact and the net health impact (Johansson
and Norheim 2011, Norheim 2013, Asaria, Griffin et al. 2015). The basic idea is to conduct sensitivity
analysis in order to find out how much you need to care about reducing inequality to recommend a
fairer option rather than a health maximising option. This sensitivity analysis can be done by
specifying alternative ‘equity weights’ for health benefits to different people, or alternative values of
an ‘equity parameter’ quantifying your degree of concern for health equity versus improving total
health. Box 4 illustrates, using an example comparing more equitable and more cost-effective ways
of increasing uptake of colorectal cancer screening in England.
Many different characteristics of health policies and the people affected by them could be used as
the basis for setting ‘equity weights’, and several different equity weighting systems have been
proposed (Wailoo, Tsuchiya et al. 2009, Cookson, Griffin et al. 2014). Many of these systems do not
pay special attention to the social characteristics of people, but rather focus on their health
characteristics – for example, their current severity of illness, or their overall lifetime experience of
health including past, present and future health (Nord 1993, Williams 1997, Lindemark, Norheim et
al. 2014, Ottersen, Mæstad et al. 2014, Ottersen, Forde et al. 2016, Rowen, Brazier et al. 2016).
These systems tend to be based upon one or more ‘equity parameters’, such as an inequality
aversion parameter within a social welfare function, which specifies how much you care about
reducing unfair health inequality rather than how much you care about individuals with certain
social characteristics (Asaria, Griffin et al. 2013). An ‘equity parameter’ then indirectly implies a set
of equity weights for people with different characteristics, in conjunction with information about the
existing distribution of health between individuals with different characteristics. These implied
weights will then change in response to changes in the distribution of health and social variables.
If decision makers are interested in impacts on relative inequality (e.g. life expectancy ratios), as well
as absolute inequality (e.g. life expectancy gaps) then they will require an estimate of the baseline
distribution of health – since the impact on relative inequality depends upon the baseline.
Furthermore, since equity trade-off analysis requires the estimation of net health impacts, benefits
have to be quantified using the same metric as opportunity costs, for example using annual
mortality risk, life years, QALYs or DALYs.
20 CHE Research Paper 135
Box 4 – Example of an equity weighting analysis
Bowel Cancer Screening in England – a “distributional cost-effectiveness analysis” (DCEA)
Quality Adjusted Life Expectancy
This study compares two strategies for increasing uptake of a universal bowel cancer screening programme.
The “targeted” strategy focuses on social groups with low uptake, by sending a personalised reminder from
their family doctor; the “universal” strategy sends a generic reminder to everyone. Inequality impacts are
analysed by deprivation, ethnicity and gender, and then combined to assess the overall impact. The
illustrations below focus on inequality by deprivation. The left hand panel below shows unequal health by
deprivation group before the screening programme. The right hand panel shows unequal uptake of
screening from the “standard” programme and after the “targeted” and “universal” reminder strategies.
75
71.79
70.43
70
73.50
61.70
60
55
50
Most
Deprived
IMD 4
targeted
universal
80%
75%
70%
65%
60%
55%
50%
45%
40%
67.87
65
standard
IMD 3
IMD 2
Most
Deprived
Least
Deprived
Unequal health: Quality Adjusted Life Expectancy
at birth by deprivation group
IMD 4
IMD 3
IMD 2
Least
Deprived
Unequal screening uptake before and after the
two reminder strategies
The left hand panel below shows the resulting changes in health. The targeted strategy is inequality
reducing but produces less total health than the universal strategy, which is inequality increasing. Which
strategy you consider best depends on how much you care about reducing health inequality versus
improving total health. The right hand panel shows how this trade off can be quantified using the
inequality aversion parameter from an Atkinson social welfare function. Which strategy you consider better
depends on how much you care about reducing health inequality – in this case, the “targeted” strategy is
better if your inequality aversion parameter is greater than 8. Recent survey data suggest the average
member of the English general public has an inequality aversion parameter of around 11 (Robson, Asaria et
al. 2016).
1500
targeted
universal
1000
Universal EDE - Targeted EDE
(Population QALYs)
Incremental Per Person QALYs
Compared to No Intervention
0.005
0.004
0.003
0.002
0.001
0.000
Universal
Better
500
0
-500
-1000
Targeted
Better
-1500
-2000
-0.001
-2500
Most
Deprived
IMD 4
IMD 3
IMD 2
Least
Deprived
Change in health due to reminders
0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40
Constant Relative Inequality Aversion (Atkinson ε)
How the best strategy depends on the inequality
aversion parameter
Reference Asaria, Miqdad, Susan Griffin, and Richard Cookson. "Distributional Cost-Effectiveness Analysis: A
Tutorial." Medical Decision Making 36.1 (2016): 8-19.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 21
In the past, some authors have taken a sceptical view about the value of equity weighting analysis
(Sassi, Archard et al. 2001). However, we take a more pragmatic approach. Equity weighting analysis
will not be useful when there are no trade-offs between improving total health and health equity.
However, equity trade-off analysis may be useful when decision makers do face difficult trade-offs.
Often there will not be trade-offs, as seen in many examples where there are ‘win-win’ scenarios.
However, previous studies have often not paid close attention to the question of whether additional
delivery costs are required in disadvantaged communities due to poor infrastructure, weak
governance and access barriers; nor have they analysed the distribution of opportunity costs from
public or donor budgets. Due to issues of this kind, there will sometimes be difficult trade-offs due
to social variation in access, adherence and effectiveness, and social variation in opportunity costs,
and in such cases there may be a need for equity trade-off analysis (Eyal, Hurst et al. 2013). Analysis
of these trade-offs can help decision makers understand the implications of alternative value
judgements; and the use of equity parameters and weights can help form useful ‘benchmarks’ for
decision makers to compare across different decisions. As with the cost-effectiveness threshold, this
kind of benchmark can be useful to decision makers as a way of justifying their decisions in the face
of competing political demands from rival stakeholders, and of facilitating both transparency and
consistency between decisions.
22 CHE Research Paper 135
Conclusion
In response to growing policy concern about health equity, the tools of economic evaluation are
being re-fashioned to provide useful evidence about health equity impacts and trade-offs. Our
report has described the key concepts and methods now available for using cost-effectiveness
analysis to analyse health equity concerns. We have distinguished three general approaches: (1)
equity impact analysis, (2) equity constraint analysis and (3) equity weighting analysis. We have also
described two specific methods of ‘distributional economic evaluation’ that incorporate equity
impact analysis within economic evaluation: ‘extended cost-effectiveness analysis’ (ECEA) and
‘distributional cost-effectiveness analysis’ (DCEA). ECEA examines the distribution of costs, health
outcomes and financial risk protection outcomes; and DCEA combines equity impact analysis with
equity constraint and trade-off analyses.
Our report has the following conclusions for decision makers who use cost-effectiveness studies:



Some decisions involve equity trade-offs between total health and health equity.
Equity trade-offs can occur when delivering services to more disadvantaged
communities if it requires additional investment of resources, due to poor
infrastructure, weak governance, and utilisation barriers.
In the face of equity trade-offs, equity-informative economic evaluation can be used to
provide useful quantitative information to help make fairer decisions and improve
health for all.
Our report also has the following conclusions for analysts who produce cost-effectiveness studies:




Who gains most depends on social variation in several factors including health risks,
access to care, adherence to care, quality of care, clinical effects, and capacity to
benefit.
Providing a full picture of health equity impacts requires analysing not only who gains
but also who bears the opportunity costs of diverting scarce resources from other uses.
Accurate estimates of health equity impacts requires analysis of social inequality in
service delivery and the implementation costs of attempting to reduce this inequality,
rather than merely assuming that services will be delivered and used in the same way
and at the same cost in advantaged and disadvantaged communities
Health equity findings should include sensitivity analysis that helps decision makers
understand the implications of alternative value judgements about equity.
Using CEA to analyse distributional equity impacts and trade-offs requires the same basic analytical
skills as standard cost-effectiveness analysis. However, it is more demanding in terms of data
requirements because it requires social distributions of key parameters rather than merely
population average values. Data limitations can be particularly severe in low-income countries which
lack basic health information systems such as vital statistics on births and deaths, and hospital and
primary care administrative data. However, as the DCP3 project has amply demonstrated,
distributional economic evaluation studies can successfully be performed in low-income countries
using existing datasets and existing, large survey-based datasets in low-income countries such as
UNICEF’s Multiple Indicator Cluster Survey (MICS) and ICF International’s Demographic and Health
Surveys (DHS), among others, which include information disaggregated by socioeconomic status
(ICF-International-USAID 2016, UNICEF 2016). Datasets are improving all the time, and in the future
it will become increasingly feasible to estimate social distributions of key parameters in economic
evaluation studies.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 23
Three important frontiers of research in this field are (1) research on the distribution of opportunity
costs of funding health programmes from different sources (e.g. public health care, donor budgets,
taxation), (2) research on the implementation costs of achieving more equal delivery and utilisation
of services in disadvantaged communities, and (3) research on public and stakeholder views about
health equity trade-offs to provide benchmark values to help guide decision makers faced with
difficult policy trade-offs. Work on the distribution of opportunity costs issue is ongoing in England,
drawing on existing research on variation in health care expenditure and outcomes at sub-national
level (Claxton, Martin et al. 2013). This issue is particularly challenging in low- and middle-income
countries where the existence of informal health systems, donor aid, and a variety of health
financing models creates challenges for identifying where the opportunity costs lie, and is a critical
area for further research.
We hope this report will help those who produce, commission and use evidence to support priority
setting in health care and public health to navigate the practical options for using the techniques of
cost-effectivenes analysis to provide policy makers with more useful information about the health
equity implications of their decisions.
24 CHE Research Paper 135
Further reading
Asada, Y., J. Hurley, O. F. Norheim and M. Johri (2014). A three-stage approach to measuring health
inequalities and inequities. International Journal for Equity in Health 13(1): 98.
Asaria, Miqdad, Susan Griffin, and Richard Cookson. ‘Distributional Cost-Effectiveness Analysis: A
Tutorial.’ Medical Decision Making 36.1 (2016): 8-19.
Cookson, R., M. Drummond and H. Weatherly (2009). ‘Explicit incorporation of equity considerations
into economic evaluation of public health interventions.’ Health Econ Policy Law 4(Pt 2): 231-245.
Cleary, S., Mooney, G., & McIntyre, D. (2010). Equity and efficiency in HIV-treatment in South Africa:
the contribution of mathematical programming to priority setting. Health Economics, 19(10), 11661180. doi:10.1002/hec.1542
Culyer, A. J. (2012). ‘Hic Sunt Dracones: The Future of Health Technology Assessment—One
Economist’s Perspective.’ Medical Decision Making 32(1): E25-E32.
Culyer, A. J. (2016). ‘HTA – Algorithm or Process?; Comment on ‘Expanded HTA: Enhancing Fairness
and Legitimacy’.’ International Journal of Health Policy and Management 5(8): 501-505.
Culyer AJ, Bombard Y. An equity framework for health technology assessments. Med Dec Making.
2012; 32: 428 – 441.
Daniels, N., T. Porteny and J. Urritia (2016). ‘Expanded HTA: enhancing fairness and legitimacy.’
International journal of health policy and management 5(1): 1-3.
Johri, M. and O. Norheim (2012). ‘Can cost-effectiveness analysis integrate concerns for equity?
Systematic review.’ Int J Technol Assess Health Care 28(2): 125 - 132.
Kjellsson, G., U.-G. Gerdtham and D. Petrie (2015). ‘Lies, damned lies, and health inequality
measurements: understanding the value judgments.’ Epidemiology 26(5): 673-680.
Love-Koh, J., M. Asaria, R. Cookson and S. Griffin (2015). ‘The Social Distribution of Health:
Estimating Quality-Adjusted Life Expectancy in England.’ Value Health 18(5): 655-662
Norheim, O., R. Baltussen, M. Johri, D. Chisholm, E. Nord, D. Brock, P. Carlsson, R. Cookson, N.
Daniels, M. Danis, M. Fleurbaey, K. Johansson, L. Kapiriri, P. Littlejohns, T. Mbeeli, K. Rao, T. T.-T.
Edejer and D. Wikler (2014). ‘Guidance on priority setting in health care (GPS-Health): the inclusion
of equity criteria not captured by cost-effectiveness analysis.’ Cost Effectiveness and Resource
Allocation 12(1): 18.
Thokaka P, Devlin, N, Marsh K, et al. Multiple Criteria Decision Analysis for Health Care Decision
Making—An Introduction: Report 1 of the ISPOR MCDA Emerging Good Practices Task Force. Value
Health 2016;19:1-13.
Verguet, S., J. J. Kim and D. T. Jamison (2016). ‘Extended cost-effectiveness analysis for health policy
assessment: a tutorial.’ PharmacoEconomics 34(9): 913-923.
Williams, A. H. and R. A. Cookson (2006). ‘Equity-efficiency trade-offs in health technology
assessment.’ International Journal of Technology Assessment in Health Care 22(1): 1-9.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 25
Annex: Computation of the distribution of healthy life expectancy in Ethiopia
In England, the baseline socioeconomic distribution of healthy life expectancy can be estimated by
linking whole-population administrative data on mortality, population and small area deprivation,
and making adjustments for self-reported health-related quality of life based on survey data from a
nationally representative random sample (Asaria et al., 2015, Love-Koh et al., 2015). However,
estimating the distribution of healthy life expectancy is more challenging in low income countries
such as Ethiopia, which lack reliable census and vital registration systems that can provide wholepopulation data on all-age population and mortality. We developed an approach which draws on
WHO data on healthy life expectancy, a previous analysis of Demographic Health Surveillance (DHS)
survey data on child mortality by household wealth asset group, and further adjustments for the
distribution of morbidity based on additional DHS data on child and adult disease prevalence.
Our approach involved a three stage process. First, we took the average HALE values for men and
women from the WHO database (WHO, 2015). Second, relative mortality weights by socioeconomic
group were applied, based on a previous analysis of modelled life expectancy drawing on DHS data
on child mortality by household asset group. Third, relative morbidity weights by socioeconomic
group were applied based on available prevalence data. This is explained more fully below and in
Annex Table 1. Like many methods used in practice in the global health field, our approach involves
making a number of assumptions in order to extract useful information from imperfect data. It
nevertheless provides a starting point for analysis until better data becomes available.
Stage 1: Healthy Life Expectancy
Healthy life expectancy (HALE) is a summary measure of population health measured in healthy life
years (HALYs) that has begun to be included in WHO datasets of life expectancy. It is developed
using Sullivan’s method applied to country life tables and age-sex-specific estimates of severityadjusted equivalent years of healthy life lost as a fraction of total years lived by each age-sex group.
The latter is calculated by summing years of healthy life lost due to disability (YLD) across a
comprehensive set of disease and injury causes drawing on analyses from the Global Burden of
Disease study (WHO, 2014). HALE was chosen as the starting point for the analysis as it uses more
detailed data on morbidity than rival approaches such as disability-free life expectancy, which only
use binary morbidity data on whether or not a person has disability. By contrast, HALE uses cardinal
data on health-related quality of life based on disease prevalence and public views about health loss
associated with different disease states (Salomon, Wang et al. 2012). It also has the added benefit
of being available for a wide range of countries, which is of particular importance when data
limitations are so prominent. Again, no disaggregated HALE data is available but the HALE values
give average measures of population health that take account of both quality and length of life, and
therefore provide a reasonable starting point.
Stage 2: Adjust HALE according to distribution of life expectancy
As no life expectancy data that is disaggregated by social characteristics is available, modelled life
expectancies by wealth quintile (taken from Tranvag, Ali et al. 2013) were used to calculate relative
weights which were applied to the average HALE values for males (54 HALYs) and females (56 HALYs)
(WHO, 2015). The modelled life expectancy values were calculated using a modified logit life table
system which requires stratified under-5 and adult mortality data. Under-five mortality data by
gender, urban-rural residence and wealth quintiles from the 2011 EDHS was used. In 2011 the EDHS
surveyed 17,817 households 31% in urban areas and 69% in rural areas, interviewing 16,515 women
and 14,110 men. Adult mortality rates by the same groups were not available and were, therefore,
calculated using adult mortality and life expectancy from the Global Burden of Disease study 2010
and weighted ratios of under-5 mortality rates for the respective groups (see Tranvag, Ali et al. 2013
for full explanation). The relative weights were calculated by assuming the modelled life expectancy
26 CHE Research Paper 135
of the middle wealth quintile, Q3, was equal to the mean life expectancy, and relative weights for
the other quintile groups were calculated accordingly. By assuming that the distribution of life
expectancy applies equally to the morbidity and mortality components of the HALE, these weights
were applied as relative adjustment factors to the average HALE values to produce a distribution of
HALE that reflects the distribution of life expectancy among the population according to their
wealth.
Stage 3: Adjust HALE according to distribution of mortality
The distribution of HALE obtained in Stage 2 reflects the distribution of life expectancy. However, a
second adjustment is required so that it also reflects the distribution of morbidity. This adjustment
was based on data on the prevalence of disease disaggregated by wealth quintile group taken from
the Ethiopia Demographic and Health Survey (Central Statistical Agency and ICF International, 2012).
Of the available diseases for which prevalence data was available, those selected to form the basis of
the adjustment were anaemia, diarrhoea and acute respiratory infection (ARI) in children. This
selection was based on the diseases that accounted for the largest burden of disease according to
the Global Burden of Disease study and was restricted to three diseases because of the limited
number of diseases and health issues for which data is available (GBD, 2010, IHME, 2015). Only the
prevalence in children was used because of the limited data on adult morbidity that was available.
Consequently, for the morbidity adjustment it was assumed that there is equal morbidity in adults as
in children. The average morbidity prevalence was calculated from the 3 diseases selected and was
then subtracted from 100 to equate to the prevalence of good health. This was then used to
calculate the relative adjustment factors for quality of life in the same way as for the adjustment for
the distribution of life expectancy – assuming that the prevalence of good health for Q3 was equal to
the average. By applying these relative adjustment factors to the HALE values obtained in Stage 1, a
HALE distribution reflecting both the distribution of morbidity and the distribution of life expectancy
was obtained. As before, this assumed that the morbidity adjustment applies equally to the
morbidity and mortality components of HALE.
The Baseline Health Distribution
Following the adjustments for quality and length of life outlined above, the groups were ordered
from least to most healthy and adjusted for the size of the group to produce a population
distribution of HALE at birth (see Figure 1 of main text).
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 27
Annex Table 1: Modelling the Baseline Health Distribution
Stage 1
Stage 2
*Average HALE
Female
56
Wealth
Quintile
*Modelled
Life
expectancy
Adjustment
Factor (LE)
Female
Male
0.88
49.35
47.58
100Average
Morbidity
Prevalence
76.5
Q1
53.4
Q2
56.2
0.93
51.93
50.08
Q3
60.6
1
56
Q4
59.9
0.99
Q5
62.5
1.03
Male
54
Stage 3
*Source: WHO, 2015
**Source: Tranvag, Ali & Norheim, 2013
Adjusted HALE
Adjustment
Factor (QoL)
Adjusted HALE
Female
Male
0.97
48.19
46.47
78
0.99
51.71
49.87
54
78.33
1
56
54
55.35
53.38
77.97
0.99
55.09
53.13
57.76
55.69
82.97
1.06
61.17
58.99
28 CHE Research Paper 135
References
Arcaya, M. C., A. L. Arcaya and S. V. Subramanian (2015). Inequalities in health: definitions, concepts,
and theories. Glob Health Action 8: 27106.
Asada, Y., J. Hurley, O. Norheim and M. Johri (2015). Unexplained health inequality - is it unfair?
International Journal for Equity in Health 14(1): 11.
Asada, Y., J. Hurley, O. F. Norheim and M. Johri (2014). A three-stage approach to measuring health
inequalities and inequities. International Journal for Equity in Health 13(1): 98.
Asada, Y., J. Hurley, O. F. Norheim and M. Johri (2015). Unexplained health inequality–is it unfair?
International Journal for Equity in Health 14(1): 11.
Asaria, M., S. Griffin and R. Cookson (2013). Measuring health inequality in the context of costeffectiveness analysis. Health and Inequality (Research on Economic Inequality, Volume 21) Emerald
Group Publishing Limited 21: 491-507.
Asaria, M., S. Griffin, R. Cookson, S. Whyte and P. Tappenden (2015). Distributional costeffectiveness analysis of health care programmes--a methodological case study of the UK Bowel
Cancer Screening Programme. Health Econ 24(6): 742-754.
Bajekal, M., S. Scholes, H. Love, N. Hawkins, M. OFlaherty, R. Raine and S. Capewell (2012). Analysing
recent socioeconomic trends in coronary heart disease mortality in England, 2000-2007: A
population modelling study. PLoS Medicine 9: 12.
Brown, T., S. Platt and A. Amos (2014). Equity impact of interventions and policies to reduce smoking
in youth: systematic review. Tob Control 23(e2): e98-105.
Central Statistical Agency [Ethiopia] and ICF International (2012). Ethiopia Demographic and Health
Survey 2011. Addis Ababa, Ethiopia and Calverton, Maryland, USA, Central Statistical Agency and ICF
International.
Claxton, K., S. Martin, M. Soares, N. Rice, E. Spackman, S. Hinde, N. Devlin, P. C. Smith and M.
Sculpher (2013). Methods for the estimation of the NICE cost effectiveness threshold. York: University
of York.
Cookson, R. (2015). Justice and the NICE approach. Journal of medical ethics 41(1): 99-102.
Cookson, R., M. Drummond and H. Weatherly (2009). Explicit incorporation of equity considerations
into economic evaluation of public health interventions. Health Econ Policy Law 4(Pt 2): 231-245.
Cookson, R., S. Griffin and E. Nord (2014). Incorporation of concerns for fairness in economic
evaluation of health programs: overview. Encyclopedia of Health Economics. A. J. Culyer. San Diego,
Elsevier. 2: 27-34.
Cotlear, D., S. Nagpal, O. Smith, A. Tandon and R. Cortez (2015). Going Universal: How 24 Developing
Countries are Implementing Universal Health Coverage from the Bottom Up, World Bank
Publications.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 29
Culyer, A. J. (2006). The bogus conflict between efficiency and vertical equity. Health Economics
15(11): 1155-1158.
Culyer, A. J. (2012). Hic Sunt Dracones The Future of Health Technology Assessment—One
Economists Perspective. Medical Decision Making 32(1): E25-E32.
Culyer, A. J. (2016). HTA – Algorithm or Process?; Comment on Expanded HTA: Enhancing Fairness
and Legitimacy. International Journal of Health Policy and Management 5(8): 501-505.
Culyer, A. J. and Y. Bombard (2012). An equity framework for health technology assessments.
Medical Decision Making 32(3): 428-441.
Culyer, A. J. and J. Lomas (2006). Deliberative processes and evidence-informed decision making in
healthcare: do they work and how might we know? Evidence & Policy: A Journal of Research, Debate
and Practice 2(3): 357-371.
Daniels, N., T. Porteny and J. Urritia (2016). Expanded HTA: enhancing fairness and legitimacy.
International Journal of Health Policy and Management 5(1): 1-3.
Dawkins, B. (2015). Distributional Cost-Effectiveness Analysis in Global Health. Masters of Science in
Health Economics, University of York.
Drummond, M. F., M. Sculpher, K. Claxton, G. L. Stoddart and G. W. Torrance (2015). Methods for
the Economic Evaluation of Health Care Programmes, Oxford University Press.
Earnshaw, S. R., A. Richter, S. W. Sorensen, T. J. Hoerger, K. A. Hicks, M. Engelgau, T. Thompson, K.
M. Narayan, D. F. Williamson, E. Gregg and P. Zhang (2002). Optimal allocation of resources across
four interventions for type 2 diabetes. Med Decis Making 22(5 Suppl): S80-91.
Epstein, D. M., Z. Chalabi, K. Claxton and M. Sculpher (2007). Efficiency, Equity, and Budgetary
Policies Informing Decisions Using Mathematical Programming. Medical Decision Making 27(2): 128137.
Eyal, N., S. A. Hurst, O. F. Norheim and D. Wikler (2013). Inequalities in Health: Concepts, Measures,
and Ethics, Oxford University Press.
Fleurbaey, M. and E. Schokkaert (2011). Chapter Sixteen - Equity in Health and Health Care.
Handbook of Health Economics. T. G. M. Mark V. Pauly and P. B. Pedro, Elsevier. Volume 2: 10031092.
Global Burden of Disease (GBD). (2010). GBD Profile: Ethiopia. Retrieved 22 July, 2015, from
http://www.healthdata.org/sites/default/files/files/country_profiles/GBD/ihme_gbd_country_repor
t_ethiopia.pdf.
Gwatkin, D. R., A. Bhuiya and C. G. Victora (2004). Making health systems more equitable. Lancet
364(9441): 1273-1280.
Gwatkin, D. R. and A. Ergo (2011). Universal health coverage: friend or foe of health equity? The
Lancet 377(9784): 2160-2161.
30 CHE Research Paper 135
Harper, S., N. B. King, S. C. Meersman, M. E. Reichman, N. Breen and J. Lynch (2010). Implicit value
judgments in the measurement of health inequalities. Milbank Q 88(1): 4-29.
Hausman, D. (2013). Egalitarian Critiques of Health Inequalities. Inequalities in Health: Concepts,
Measures, and Ethics. N. Eyal, S. A. Hurst, O. F. Norheim and D. Wikler. Oxford, Oxford Press 27.
Holmes, J., Y. Meng, P. S. Meier, A. Brennan, C. Angus, A. Campbell-Burton, Y. Guo, D. Hill-McManus
and R. C. Purshouse (2014). Effects of minimum unit pricing for alcohol on different income and
socioeconomic groups: a modelling study. Lancet 383(9929): 1655-1664.
ICF-International-USAID. (2016). The DHS Program: Demographic and Health Surveys. Retrieved 14
July 2016, 2016, from http://www.dhsprogram.com/.
Institute for Health Metrics and Evaluation (IHME). (2015). Ethiopia. Retrieved 22 July, 2015, from
http://www.healthdata.org/ethiopia.
Jamison, D. T., L. H. Summers, G. Alleyne, K. J. Arrow, S. Berkley, A. Binagwaho, F. Bustreo, D. Evans,
R. G. Feachem and J. Frenk (2013). Global health 2035: a world converging within a generation. The
Lancet 382(9908): 1898-1955.
Johansson, K. A. and O. F. Norheim (2011). Problems with prioritization: exploring ethical solutions
to inequalities in HIV care. Am J Bioeth 11(12): 32-40.
Johri, M. and O. Norheim (2012). Can cost-effectiveness analysis integrate concerns for equity?
Systematic review. Int J Technol Assess Health Care 28(2): 125 - 132.
Johri, M. and O. F. Norheim (2012). Can cost-effectiveness analysis integrate concerns for equity?
Systematic review. Int J Technol Assess Health Care 28(2): 125-132.
Kjellsson, G., U.-G. Gerdtham and D. Petrie (2015). Lies, damned lies, and health inequality
measurements: understanding the value judgments. Epidemiology 26(5): 673-680.
Kruse, I., M. Pradhan and R. Sparrow (2012). Marginal benefit incidence of public health spending:
Evidence from Indonesian sub-national data. Journal of Health Economics 31(1): 147-157.
Lindemark, F., O. F. Norheim and K. A. Johansson (2014). Making use of equity sensitive QALYs: a
case study on identifying the worse off across diseases. Cost Effectiveness and Resource Allocation
12(1): 1.
Love-Koh, J., M. Asaria, R. Cookson and S. Griffin (2015). The Social Distribution of Health: Estimating
Quality-Adjusted Life Expectancy in England. Value Health 18(5): 655-662.
Marmot, M., J. Allen, R. Bell, E. Bloomer and P. Goldblatt (2012). WHO European review of social
determinants of health and the health divide. The Lancet 380(9846): 1011-1029.
Marmot, M., S. Friel, R. Bell, T. Houweling and S. Taylor (2008). Closing the gap in a generation:
health equity through action on the social determinants of health. Lancet 372(9650): 1661 - 1669.
McAuley, A., C. Denny, M. Taulbut, R. Mitchell, C. Fischbacher, B. Graham, I. Grant, P. OHagan, D.
McAllister and G. McCartney (2016). Informing Investment to Reduce Inequalities: A Modelling
Approach. PLoS One 11(8): e0159256.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 31
Miljeteig, I., K. A. Johansson, S. A. Sayeed and O. F. Norheim (2010). End-of-life decisions as bedside
rationing. An ethical analysis of life support restrictions in an Indian neonatal unit. J Med Ethics
36(8): 473-478.
Noor, A. M., A. A. Amin, W. S. Akhwale and R. W. Snow (2007). Increasing coverage and decreasing
inequity in insecticide-treated bed net use among rural Kenyan children. PLoS Med 4(8): e255.
Nord, E. (1993). The trade-off between severity of illness and treatment effect in cost-value analysis
of health care. Health Policy 24(3): 227-238.
Norheim, O., R. Baltussen, M. Johri, D. Chisholm, E. Nord, D. Brock, P. Carlsson, R. Cookson, N.
Daniels, M. Danis, M. Fleurbaey, K. Johansson, L. Kapiriri, P. Littlejohns, T. Mbeeli, K. Rao, T. T.-T.
Edejer and D. Wikler (2014). Guidance on priority setting in health care (GPS-Health): the inclusion of
equity criteria not captured by cost-effectiveness analysis. Cost Effectiveness and Resource Allocation
12(1): 18.
Norheim, O., T. Ottersen, F. Berhane, B. Chitah, R. Cookson, N. Daniels, N. Eyal, W. Flores, A.
Gosseries and D. Hausman (2014). Making fair choices on the path to universal health coverage:
Final report of the WHO consultative group on equity and universal health coverage, World Health
Organization.
Norheim, O. F. (2010). Priority to the young or to those with least lifetime health? Am J Bioeth 10(4):
60-61.
Norheim, O. F. (2013). Atkinsons Index Applied to Health. Inequalities in Health: Concepts,
Measures, and Ethics. N. Eyal, S. A. Hurst, O. F. Norheim and D. Wikler. Oxford, Oxford Scholarship
Online: 30.
Norheim, O. F., R. Baltussen, M. Johri, D. Chisholm, E. Nord, D. Brock, P. Carlsson, R. Cookson, N.
Daniels, M. Danis, M. Fleurbaey, K. A. Johansson, L. Kapiriri, P. Littlejohns, T. Mbeeli, K. D. Rao, T. T.
Edejer and D. Wikler (2014). Guidance on priority setting in health care (GPS-Health): the inclusion of
equity criteria not captured by cost-effectiveness analysis. Cost Eff Resour Alloc 12: 18.
ODonnell, O., E. v. van Doorslaer, A. Wagstaff and M. Lindelow (2008). Analyzing health equity using
household survey data : a guide to techniques and their implementation. Washington, D.C., World
Bank.
ONeill, J., H. Tabish, V. Welch, M. Petticrew, K. Pottie, M. Clarke, T. Evans, J. Pardo Pardo, E. Waters,
H. White and P. Tugwell (2014). Applying an equity lens to interventions: using PROGRESS ensures
consideration of socially stratifying factors to illuminate inequities in health. J Clin Epidemiol 67(1):
56-64.
Ottersen, T., R. Forde, M. Kakad, A. Kjellevold, H. O. Melberg, A. Moen, A. Ringard and O. F. Norheim
(2016). A new proposal for priority setting in Norway: Open and fair. Health Policy.
Ottersen, T., O. Mæstad and O. F. Norheim (2014). Lifetime QALY prioritarianism in priority setting:
quantification of the inherent trade-off. Cost Effectiveness and Resource Allocation 12(1): 1.
Robberstad, B. and O. F. Norheim (2011). Incorporating concerns for equal lifetime health in
evaluations of public health programs. Soc Sci Med 72(10): 1711-1716.
32 CHE Research Paper 135
Robson, M., M. Asaria, A. Tsuchiya, S. Ali and R. A. Cookson (2016). Eliciting the Level of Health
Inequality Aversion in England, Centre for Health Economics, University of York: 1-19.
Rowen, D., J. Brazier, C. Mukuria, A. Keetharuth, A. R. Hole, A. Tsuchiya, S. Whyte and P. Shackley
(2016). Eliciting societal preferences for weighting QALYs for burden of illness and end of life.
Medical Decision Making 36(2): 210-222.
Salomon, J. A., H. Wang, M. K. Freeman, T. Vos, A. D. Flaxman, A. D. Lopez and C. J. Murray (2012).
Healthy life expectancy for 187 countries, 1990-2010: a systematic analysis for the Global Burden
Disease Study 2010. Lancet 380(9859): 2144-2162.
Sassi, F., L. Archard and J. Le Grand (2001). Equity and the economic evaluation of healthcare. Health
Technol Assess 5(3): 1-138.
Shah, K. K., R. Cookson, A. J. Culyer and P. Littlejohns (2013). NICEs social value judgements about
equity in health and health care. Health Econ Policy Law 8(2): 145-165.
Tranvag, E. J., M. Ali and O. F. Norheim (2013). Health inequalities in Ethiopia: modeling inequalities
in length of life within and between population groups. Int J Equity Health 12: 52.
Tugwell, P., D. de Savigny, G. Hawker and V. Robinson (2006). Applying clinical epidemiological
methods to health equity: the equity effectiveness loop. BMJ 332(7537): 358-361.
UNICEF. (2016). Multiple Indicator Cluster Surveys (MICS). Retrieved 14 July 2016, 2016, from
http://mics.unicef.org/.
Verguet, S. and D. T. Jamison (forthcoming). Applications of extended cost-effectiveness analysis
(ECEA) methodology in DCP3. Disease Control Priorities. D. T. Jamison, R. A. Nugent, H. Gelband et al.
Washington, DC, World Bank. 9.
Verguet, S., J. J. Kim and D. T. Jamison (2016). Extended cost-effectiveness analysis for health policy
assessment: a tutorial. PharmacoEconomics 34(9): 913-923.
Verguet, S., R. Laxminarayan and D. T. Jamison (2015). Universal public finance of tuberculosis
treatment in India: an extended cost-effectiveness analysis. Health Econ 24(3): 318-332.
Verguet, S., S. Murphy, B. Anderson, K. A. Johansson, R. Glass and R. Rheingans (2013). Public
finance of rotavirus vaccination in India and Ethiopia: an extended cost-effectiveness analysis.
Vaccine 31(42): 4902-4910.
Wagstaff, A. (2015). Commentary: Value Judgments in Health Inequality Measurement.
Epidemiology (Cambridge, Mass.) 26(5): 670.
Wagstaff, A., D. Cotlear, P. H.-V. Eozenou and L. R. Buisman (2016). Measuring progress towards
universal health coverage: with an application to 24 developing countries. Oxford Review of
Economic Policy 32(1): 147-189.
Wailoo, A., A. Tsuchiya and C. McCabe (2009). Weighting must wait. Pharmacoeconomics 27(12):
983-989.
Fairer decisions, better health for all:health equity and cost-effectiveness analysis 33
Whitehead, M. (1992). The concepts and principles of equity and health. International journal of
health services 22: 429-445.
WHO (2014). Making Fair Choices on the Path to Universal Health Coverage: Final Report of the
WHO Consultative Group on Equity and Universal Health Coverage, World Health Organization: 84.
Wilkinson, R. and K. Pickett (2010). The spirit level. Why equality is better for.
Williams, A. (1997). Intergenerational equity: an exploration of the fair innings argument. Health
Econ 6(2): 117-132.
Williams, A. H. and R. A. Cookson (2006). Equity-efficiency trade-offs in health technology
assessment. International Journal of Technology Assessment in Health Care 22(1): 1-9.
World Health Organisation (2014). WHO methods for life expectancy and healthy life expectancy.
Global Helath Estimates Technical Paper WHO/HIS/HSI/GHE/2014.5. Geneva, Department of Health
Statistics and Information Systems, WHO.
World Health Organisation (2015). Life Expectacy Data by Country.
World Health Organization (2013). Handbook on health inequality monitoring with a special focus on
low-and middle-income countries, World Health Organization.
World Health Organization (2015). Tracking universal health coverage: first global monitoring report,
World Health Organization.